The Accuracy of Acquiring Heart Rate Variability from Portable Devices: A Systematic Review and Meta-Analysis.

BACKGROUND: Advancements in wearable technology have provided practitioners and researchers with the ability to conveniently measure various health and/or fitness indices. Specifically, portable devices have been devised for convenient recordings of heart rate variability (HRV). Yet, their accuracies remain questionable.

OBJECTIVE: The aim was to quantify the accuracy of portable devices compared to electrocardiography (ECG) for measuring a multitude of HRV metrics and to identify potential moderators of this effect.

METHODS: This meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. Articles published before July 29, 2017 were located via four electronic databases using a combination of the terms related to HRV and validity. Separate effect sizes (ESs), defined as the absolute standardized difference between the HRV value recorded using the portable device compared to ECG, were generated for each HRV metric (ten metrics analyzed in total). A multivariate, multi-level model, incorporating random-effects assumptions, was utilized to quantify the mean ES and 95% confidence interval (CI) and explore potential moderators.

RESULTS: Twenty-three studies yielded 301 effects and revealed that HRV measurements acquired from portable devices differed from those obtained from ECG (ES = 0.23, 95% CI 0.05-0.42), although this effect was small and highly heterogeneous (I2  = 78.6%, 95% CI 76.2-80.7). Moderator analysis revealed that HRV metric (p <0.001), position (p = 0.033), and biological sex (β = 0.45, 95% CI 0.30-0.61; p <0.001), but not portable device, modulated the degree of absolute error. Within metric, absolute error was significantly higher when expressed as standard deviation of all normal-normal (R-R) intervals (SDNN) (ES = 0.44) compared to any other metric, but was no longer significantly different after a sensitivity analysis removed outliers. Likewise, the error associated with the tilt/recovery position was significantly higher than any other position and remained significantly different without outliers in the model.

CONCLUSIONS: Our results suggest that HRV measurements acquired using portable devices demonstrate a small amount of absolute error when compared to ECG. However, this small error is acceptable when considering the improved practicality and compliance of HRV measurements acquired through portable devices in the field setting. Practitioners and researchers should consider the cost-benefit along with the simplicity of the measurement when attempting to increase compliance in acquiring HRV measurements.